No Arabic abstract
Entanglement, one of the central mysteries of quantum mechanics, plays an essential role in numerous applications of quantum information theory. A natural question of both theoretical and experimental importance is whether universal entanglement detection is possible without full state tomography. In this work, we prove a no-go theorem that rules out this possibility for any non-adaptive schemes that employ single-copy measurements only. We also examine in detail a previously implemented experiment, which claimed to detect entanglement of two-qubit states via adaptive single-copy measurements without full state tomography. By performing the experiment and analyzing the data, we demonstrate that the information gathered is indeed sufficient to reconstruct the state. These results reveal a fundamental limit for single-copy measurements in entanglement detection, and provides a general framework to study the detection of other interesting properties of quantum states, such as the positivity of partial transpose and the $k$-symmetric extendibility.
The phenomenon of quantum entanglement marks one of the furthest departures from classical physics and is indispensable for quantum information processing. Despite its fundamental importance, the distribution of entanglement over long distances trough photons is unfortunately hindered by unavoidable decoherence effects. Entanglement distillation is a means of restoring the quality of such diluted entanglement by concentrating it into a pair of qubits. Conventionally, this would be done by distributing multiple photon pairs and distilling the entanglement into a single pair. Here, we turn around this paradigm by utilising pairs of single photons entangled in multiple degrees of freedom. Specifically, we make use of the polarisation and the energy-time domain of photons, both of which are extensively field-tested. We experimentally chart the domain of distillable states and achieve relative fidelity gains up to 13.8 %. Compared to the two-copy scheme, the distillation rate of our single-copy scheme is several orders of magnitude higher, paving the way towards high-capacity and noise-resilient quantum networks.
In this paper, we continue our investigation on controlling the state of a quantum harmonic oscillator, by coupling it to a reservoir composed of a sequence of qubits. Specifically, we show that sending qubits separable from each other but initialised at different states in pairs can stabilise the oscillator at squeezed states. However, only if entanglement is allowed in the reservoir qubit can we stabilise the oscillator at a wider set of squeezed states. This thus provides a proof for the necessity of involving entanglement in the reservoir qubits input to the oscillator, as regard to the stabilisation of quantum states in the proposed system setting. On the other hand, this system setup can be in turn used to estimate the coupling strength between the oscillator and reservoir qubits. We further demonstrate that entanglement in the reservoir input qubits contributes to the corresponding quantum Fisher information. From this point of view, entanglement is proved to play an indispensable role in the improvement of estimation precision in quantum metrology.
Entanglement shared among multiple parties presents complex challenges for the characterisation of different types of entanglement. One of the most basic insights is the fact that some mixed states can feature entanglement across every possible cut of a multipartite system, yet can be produced via a mixture of partially separable states. To distinguish states that genuinely cannot be produced from mixing partially separable states, the term genuine multipartite entanglement was coined. All these considerations originate in a paradigm where only a single copy of the state is distributed and locally acted upon. In contrast, advances in quantum technologies prompt the question of how this picture changes when multiple copies of the same state become locally accessible. Here we show that multiple copies unlock genuine multipartite entanglement from partially separable states, even from undistillable ensembles, and even more than two copies can be required to observe this effect. With these findings, we characterise the notion of partial separability in the paradigm of multiple copies and conjecture a strict hierarchy of activatable states and an asymptotic collapse of hierarchy.
We present an experimentally feasible and efficient method for detecting entangled states with measurements that extend naturally to a tomographically complete set. Our detection criterion is based on measurements from subsets of a quantum 2-design, e.g., mutually unbiased bases or symmetric informationally complete states, and has several advantages over standard entanglement witnesses. First, as more detectors in the measurement are applied, there is a higher chance of witnessing a larger set of entangled states, in such a way that the measurement setting converges to a complete setup for quantum state tomography. Secondly, our method is twice as effective as standard witnesses in the sense that both upper and lower bounds can be derived. Thirdly, the scheme can be readily applied to measurement-device-independent scenarios.
This paper revisits human-object interaction (HOI) recognition at image level without using supervisions of object location and human pose. We name it detection-free HOI recognition, in contrast to the existing detection-supervised approaches which rely on object and keypoint detections to achieve state of the art. With our method, not only the detection supervision is evitable, but superior performance can be achieved by properly using image-text pre-training (such as CLIP) and the proposed Log-Sum-Exp Sign (LSE-Sign) loss function. Specifically, using text embeddings of class labels to initialize the linear classifier is essential for leveraging the CLIP pre-trained image encoder. In addition, LSE-Sign loss facilitates learning from multiple labels on an imbalanced dataset by normalizing gradients over all classes in a softmax format. Surprisingly, our detection-free solution achieves 60.5 mAP on the HICO dataset, outperforming the detection-supervised state of the art by 13.4 mAP